Faster unfolding speed in mutant HγD-Crys proteins than the wild-type

We first study the overall unfolding process of the wild-type and all the mutant proteins to access the influence of the various site-specific mutations. We also compare the unfolding speed of each individual domain. The unfolding process can be illustrated via plotting the fraction of native contacts Q of each domain with respect to the simulation time. Here, a native contact between two residues is counted if any heavy atom of residue i is within 6.5 Å of any heavy atom of another residue j (j – i > 3) in the crystal structure of the protein. For the folded state, Q ≈ 1, and for the fully stretched state, Q ≈ 0, thereby the time evolution of Q value for each domain reflects its corresponding unfolding dynamics.

As shown in Fig. 2, for all the mutants, the N-td unfolds faster than that in the wild-type. For the wild-type, the timescale for Q(N-td) to decrease to < 0.3 (N-td loses most of its structures) is ~150 ns. Whereas, for the Y45A/Y50A_N-td, Y133A/Y138A_C-td and 4Y>4A mutant proteins, it takes ~70, ~50, and ~40 ns for Q(N-td) to decrease to < 0.3, respectively, which is much quicker than that in the wild-type. Different trajectories show consistent results (more below). Meanwhile, we observe that the effect of the N-td substitutions (Y45A/Y50A_N-td mutant) on the unfolding of the C-td is negligible (C-td remains largely intact). However, for the Y133A/Y138A_C-td and 4Y>4A mutants, the C-td also unfolds much faster as compared to the wild-type and Y45A/Y50A_N-td mutant (see Fig. 2 C and D red curve).

Open in a separate window

Figure 2

The time evolution of the fraction of native contacts Q for one representative runs of the wild-type (A), and mutants Y45A/Y50A_N-td (B), Y133A/Y138A_C-td (C) and 4Y>4A (D), respectively. See the main text for the definition of the native contact between any two residues.

Fig. 3 shows another perspective of the fraction of native contacts by plotting Q(N-td) and Q(C-td) against each other, for all five independent simulation trajectories of each system. The results indicate that for each simulated system (wild-type HγD-Crys or mutants), the unfolding pathways of the N-td or C-td for all five independent trajectories are very similar, though the exact time for each event could be slightly different. Overall, our current simulation results clearly show that: the Y45A/Y50A_N-td mutant accelerates the unfolding of the N-td, but keeps the C-td intact, while the Y133A/Y138A_C-td mutant not only accelerates the unfolding of the N-td, but also the C-td. As for the 4Y>4A mutant, it results in additional disruption to both the C-td and N-td, with an even faster unfolding speed than the Y133A/Y138A_C-td mutant. Moreover, we observe a clear “two-stage” unfolding process for the wild-type and all mutants (with the N-td unfolds first, followed by the C-td) (see Figure 2 and ​and3),3), which implies the formation of partially unfolded intermediate structures with intact C-td and unfolded N-td. These results are fully consistent with previous site-directed mutagenesis studies.¹⁸

Open in a separate window

Figure 3

The fraction of native contacts Q for the two domains are plotted against each other, data from all five independent simulation trajectories for each type system. Each data point on plots is colored from blue to red according to its time sequence during unfolding. (A) wild-type, (B) Y45A/Y50A_N-td, (C) Y133A/Y138A_C-td, and (D) 4Y>4A. The color bar

Similar unfolding pathway for the wild-type and mutant proteins

In Fig. 4B we compare the root-mean-square fluctuation (RMSF) of each residue (represented by C_α) for the wild-type and the three mutants in the early stage of unfolding, i.e., before the β-sheets start to break up (<= 30 ns in our simulations). The RMSF results show that the loop regions are most flexible and the mutant proteins undergo much larger fluctuations than the wild-type, notably in the loop regions of the mutation sites. This is because Y45/Y50 or Y133/Y138 pair in the wild-type packs better with the hydrophobic core at the interface from being “attacked” by water than the corresponding Ala pair [bottom left figures in Fig. 1 (A–D)], and also because the local stable sandwich-like Y45-R79-Y50 or Y133-R167-Y138 cluster formed by the cation-π interactions is disrupted due to the Ala substitutions [see Fig. 1 (A–D) right column. We note that it has been reported that the CHARMM force field used here can properly model the cation-π interactions even though it is non-polarizable force field³⁶]. These results indicate that the unfolding process for the wild-type and all the mutants may start from these unstructured loop regions.

Open in a separate window

Figure 4

(A) The definition of the serial number of β-sheets within each motif of the N-td and C-td. The two motifs within each domain are divided by the black line. (B) RMSF of each residue for four types of simulated systems at the first 30 ns. The time evolution of fraction of native contacts Q of each β-sheet (C–F, top panel), the number of contacts between hydrophobic residues at the interface (C–F, middle panel), and the number of contacts between hydrophobic residues within each domain (C–F, bottom panel) for the wild-type protein (C), and mutants Y45A/Y50A_N-td (D), Y133A/Y138A_C-td (E), and 4Y>4A (F).

To clarify the differences in the unfolding behavior between the wild-type and the three mutants, we then examine the time evolution of the structures during the unfolding process for each system. Since HγD-Crys consists of mainly β-sheets (β-sheets content > 80%), we use the fraction of native contacts Q of each β-sheet to assess their respective persistence (or stability) as time progresses. In Fig. 4C–F, the top panel of each figure shows the secondary structural components of each β-sheet as function of time for the wild-type and three mutants, respectively.

For the wild-type (Fig. 4C), the major β-sheets in the C-td are fairly stable till the very end of the simulation, while those in the N-td are less stable, with both β-sheet 2 and 3 in motif 1 losing their secondary structures completely at ~ 70 and ~ 180 ns, respectively, and β-sheets in motif 2 starting to fully unfold at ~ 200 ns (Fig. 4C top). Interestingly, motif 2 shows very similar unfolding trend with the hydrophobic core at the interface formed by motifs 2 and 4 (Fig. 4C top and middle), while motif 1 shows very similar unfolding process with the hydrophobic core of the N-td (Fig. 4C top and bottom red curve). These results imply that motif 4 in the C-td contributes significantly to the stability of motif 2, and the hydrophobic core within the N-td also makes important contribution to the stability of motif 1.

Similar to the wild-type, the stability of the C-td in the Y45A/Y50A_N-td mutant (Fig. 4D), is largely unaffected. However, the stability of the N-td is noticeably weakened, especially for motif 2. We also note that in this mutant, the hydrophobic core at the interface is quickly disrupted at the beginning of the simulation and becomes mostly breached at ~ 60 ns (faster than that in the wild-type). The same is true for motif 2, which is largely destroyed after only ~40–60 ns, again faster than the wild-type. Interestingly, motif 1 and the hydrophobic core within the N-td also show a highly correlated unfolding pathway, but with their unfolding speed comparable to the wild-type. These results confirm that the hydrophobic core within the N-td and at the interface largely contribute to the stability of motif 1 and motif 2, respectively.

For the Y133A/Y138A_C-td mutant (Fig. 4E), the stability of both N-td and C-td is remarkably reduced, albeit it is more stable for the C-td than the N-td. In the N-td, motif 2 and the hydrophobic core at the interface, as well as motif 1 and the hydrophobic core within the N-td, are again found to unfold synchronously. Here, we find that the unfolding speed is much faster than both the wild-type and Y45A/Y50A_N-td mutant. In the C-td, as the hydrophobic core of the domain loses integrity gradually, both β-sheets in motifs 3 and 4 slowly disappear correspondingly as well. These results suggest that the stability of motifs 3 and 4 highly correlates with the hydrophobic core within the C-td. The higher stability of the C-td with respect to the N-td is mainly caused by C-td’s more compact hydrophobic core.

As for the 4Y>4A mutant (Fig. 4F), most of the unfolding features are very similar to those of Y133A/Y138A_C-td mutant. For example, motif 2 and the hydrophobic core at the interface unfold simultaneously; however, the unfolding speed in this mutant is even faster, indicating more disruptions to both C-td and N-td.

From these structural analyses for the wild-type and all the three mutants, we again reveal an asynchronous unfolding behavior for the N-td and C-td, namely, the N-td unfolds first, followed by the C-td. Moreover, we also notice that the Y45A/Y50A_N-td mutant only accelerates the unfolding of the N-td, whereas the Y133A/Y138A_C-td mutant accelerates both the N-td and C-td unfolding. The 4Y>4A mutant results in additional disruption to both the N-td and C-td, with even faster unfolding speed. These findings indicate that the stability of motif 2 in the N-td is heavily influenced by the interdomain hydrophobic interface, while motif 1 in the N-td, and motifs 3 and 4 in the C-td, are mainly stabilized by their respective intradomain hydrophobic cores. It is interesting to note that maintaining a dry and close packed hydrophobic core is crucial for protein stability and integrity, as shown in many previous protein folding and peptide self-assembly studies.^37–40 It has also been recently shown that the nanoscale dewetting (drying) within the hydrophobic core provides a significant driving force for the protein folding and is important for protein stability;^{28, 41–47} while the intrusion of water (or other solvents) into the protein interior usually acts as the initial step for protein unfolding.^48–54

The Tyr-to-Ala pair mutations disrupt the hydrophobic core at the interdomain interface

To further investigate how the Tyr-to-Ala pair mutations in both domains give rise to the aforementioned different unfolding behavior for the N-td and C-td, we examine the structural details of the interdomain interface of the wild-type protein. We observe that a well-defined hydrophobic core (blue wireframe in Fig. 1A left column) is shielded by a loosely connected hydrophilic “residual circle” (bold bond in Fig. 1A left column). The Y45, Y50, Y133, and Y138 residues are members belonging to this hydrophilic “residual circle”. More importantly, their bulky planar phenol side-chain can tightly contact with their neighboring hydrophobic residues, and thus act as the solvent blocker for the interdomain interfacial hydrophobic core (Fig. 5 A, and C–E). The amphiphilic nature of the Tyr side chain and its relatively large size can make favorable interactions with both polar residues/waters (outside the dry cavity) and non-polar residues (inside the dry cavity), thus smoothly bridging the polar (outside) and non-polar (inside) regions. When these pairs, Y45/Y50, or Y133/Y138, or both, are substituted with the Ala pairs, apparent water leakages to the interdomain hydrophobic core emerge due to the much smaller side chain of alanines, as shown in Fig. 5 (A, B and F–H).

Open in a separate window

Figure 5

The sectional view [detailed structural analyses] of the interdomain interface for the wild-type (A), and 4Y>4A mutant proteins (B). The overall surface is shown in light green (N-td) and light cyan (C-td), residues Y45, Y50, Y133 and Y138 are shown in red, other hydrophobic residues are shown in blue. Top view (C), side view from the direction of Y45/Y55 (D), and side view from the direction of Y133/Y138 for the wild-type protein. Top view (F), side view from the direction of A45/A50 (G), and side view from the direction of A133/A138 for the 4Y>4A mutant protein. Residues Y45, Y133, A45, and A133 are shown in blue transparent surface; residues Y50, Y138, A50, and A138 are shown in green transparent surface. Residue M43 together with V131 in the center region of hydrophobic core at the interface are shown in black; while the semi-transparent red regions are the possible water leakage areas introduced by the replacement of Tyr with Ala.

We then studied the local residual contact network of residue-sites 45, 50, 133, and 138 for both the wild-type and mutants, by calculating their local native contact probability with other residues in the first 30 ns, the period when most of the secondary structural components of the proteins are largely preserved. We observe that for the Y45A/Y50A_N-td mutant, the local contacts between A45 with L144 and M146, as well as A50 with S20, D21, H22, P23, N24, M43, S77, and C78 disappear with time. Whereas for the Y133A/Y138A_C-td mutant, the contacts between A133 with F56, as well as A138 with D107, C108, S109, C110, V131, Q164, L166, and R168 also disappear. Likewise, for the 4Y>4A mutant, the contact probabilities of A45 and A50 with their neighbor residues are lost, similar to those in Y45A/Y50A_N-td mutant, and the contact probabilities of A133 and A138 are also lost just like those in Y133A/Y138A_C-td mutant, which implies that there are additive leakages generated. These results clearly show that the functions of Y45/Y50 and Y133/Y138 pairs in protecting the well-defined interdomain hydrophobic core in the wild-type are disrupted by the Tyr-to-Ala pair(s) mutations. In the previous section, we have demonstrated that motif 2 in the N-td is mainly stabilized by the dry and well-defined hydrophobic core at the interface. Thereby, this explains why the Tyr-to-Ala pair(s) mutations can significantly increase the unfolding speed of motif 2.

Moreover, we also observe that for the wild-type, the contacts among Y50, R79, and Y45, and contacts among Y138, R167, and Y133, are very well kept. However, in the Y45A/Y50A_N-td mutant, the contact probabilities among A50, R79, and A45 decrease sharply, particularly those between A50 and R79, and A50 and A45. On the other hand, in the Y133A/Y138A_C-td mutant, the contact probabilities among A138, R167, and A133 also reduce significantly, with only the contact probability between A133 and R167 remains somewhat unchanged. In addition, the contact probabilities among Y50, R79 and Y45 also decrease largely. Finally, for the 4Y>4A mutant, we find that the contact probabilities among A50, R79, and A45, and among A138, R167, and A133 all reduce significantly. The main reason for this is the loss of the stable “sandwich-like” clusters, Y45-R79-Y50, or Y133-R167-Y138, or both, due to the alanine substitutions. These clusters in the wild-type are formed through the so-called cation-π interactions, similar to the previous findings in protein lysozyme where Arg-Trp-Arg sandwich clusters contributed significantly to the protein stability.^{24, 55} The loss of these tightly bound clusters makes the loop regions near the residue A50 or A138 or both more flexible (more discussion below). These findings are also consistent with the insights gained from above analyses of the flexibility and fluctuations (see Figure 4B).

Water “leakages” to the dry hydrophobic cores due to Tyr-to-Ala mutations

We then monitored the water solvation dynamics near the hydrophobic cores at the interdomain interface (interdomain hydrophobic core) and within each domain (intradomain hydrophobic core) of the protein, for both the wild-type and the three mutants. L44 and L132 were selected as our target residues to illustrate the dynamics of water intrusion into the intradomain hydrophobic cores of the N-td and C-td, respectively, since they are deeply buried in the very central regions of the respective hydrophobic cores (Fig. 7A). Meanwhile, residue M43 was chosen as the target for studying the penetration of water into the interdomain hydrophobic core (Fig. 7A). A water molecule is considered to be in contact with a target residue if its oxygen atom is within 4.0 Å of any heavy atom in the side chain of the target residue.

Open in a separate window

Figure 7

Locations of “target” residues M43 (black), L44 (yellow), and L132 (magenta) in the protein (A). The number of water in contact with residues M43 (B), L44 (C) and L132 (D), respectively. A contact between water and a residue is defined if oxygen atom of water is within 4.0 Å of any heavy atom in the side chain of that residue. (All data are smoothed over the near 50 data points)

Fig. 7 B–D shows the solvation dynamics of residues M43, L44, and L132, respectively, for both the wild-type and the mutants. The timescale for residue M43 to be fully hydrated is about 200, 60, 40, and 20 ns for the wild-type, Y45A/Y50A_N-td, Y133A/Y138A_C-td, and 4Y>4A mutants, respectively (Fig. 7 B). These results clearly show that Tyr-to-Ala pair(s) mutant can directly induce water (or other solvent) more readily to intrude into the interdomain hydrophobic core. Interestingly, the timescale for residue L44 (representing the intradomain hydrophobic core in N-td) to be fully hydrated is comparable with those of residue M43 for all cases (Fig. 7 C). This may be because the loss of the hydrophobic core at the interface destabilizes motif 2 of the N-td, which subsequently results in the exposure of L44 to water (it can also be seen from their close positions in sequence). This simultaneous co-occurrence of the water intrusion into the interdomain and N-td’s intradomain hydrophobic cores also explains why the N-td is less stable and unfolds once the interdomain hydrophobic core is disrupted. On the other hand, the target residue L132 in the C-td’s intradomain hydrophobic core keeps dry till the very end of the simulation in both the wild-type and Y45A/Y50A_N-td mutant. It begins to fully hydrate at about 100 and 50 ns for the Y133A/Y138A_C-td and 4Y>4A mutants, respectively. To further confirm the water solvation dynamics of these key residues M43, L44, and L132, we also analyzed their solvent accessible surface (SAS) areas as a function of simulation time (Fig. S2). Our results show that the patterns in the time evolution of these SAS areas for the WT and mutants are nearly identical to their water solvation dynamics (contacting water numbers, Figure 7). For example, the characteristic times for these residues starting to be largely exposed to water, or being fully hydrated, are nearly identical in these two plots (see Fig. S2 and Fig. 7 for details), indicating both are good metrics for investigating the wetting/drying phenomena. As further shown in supporting figures Fig. S3 C and D, at 80 and 40 ns, some apparent water leakages to the C-td’s intradomain hydrophobic core emerge near the A138 loop regions. At 120 and 60 ns, a few water molecules reach to the residue L132. And at 140 and 80 ns, residue L132 becomes fully hydrated in the Y133A/Y138A_C-td and 4Y>4A mutants, respectively. These results indicate that the replacement of Y133/Y138 with A133/A138 causes the relative stable “sandwich-like” Y133-R167-Y138 cluster to disappear, thus making the loop regions near the residue A138 more flexible. As a result, the larger fluctuations in the loop regions near A138 facilitate the intrusion of water into the intradomain hydrophobic core, which initiates the unfolding of the C-td.

We thank Peng Xiu, Xiaowei Tang, Weiqiu Zhu, Bo Zhou, and Bruce Berne for useful discussions and insightful comments. This work is partially supported by National Natural Science Foundation of China (grant Nos. 11374221, 21320102003, and 30870593), and the China Postdoctoral Science Foundation (grant No. 2012M511351). J.K. acknowledges financial support from NIH {"type":"entrez-nucleotide","attrs":{"text":"EY015834","term_id":"159080729","term_text":"EY015834"}}EY015834. R.Z. acknowledges the financial support from the IBM BlueGene Science Program. A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).